OpenArt Founder Coco Mao: Raw Products Must Also Have Positive Signals
OpenArt co-founder and CEO Coco Mao stated that as long as there is product-market fit, even in its roughest state, positive signals should be observed. She compared fit to going out without makeup: if people still come, it indicates they want the product itself, not the polish. The company's first version of the product was created in three days by a team of just three people. OpenArt currently has about 60 to 70 employees; earlier public statements indicated that around 20 people achieved over $70 million in annual recurring revenue and about 8 million monthly active users, with an average revenue of about $3.5 million per person, and claimed to be profitable. In January 2026, Canaan Partners led a $30 million Series A round, with follow-on investments from Basis Set, DCM, and others, bringing total funding to about $35 million.
Coco Mao and John Qiao, both former Google employees, founded OpenArt in 2022 after leaving Area 120. Initially, they aimed to create a "Pinterest for AI images and prompts," gaining traction after posting on Hacker News. The subscription model is based on points: approximately $7/month for 4,000 points, $14.5/month for 12,000 points, and $28/month for 24,000 points. Early users included enthusiasts and small businesses creating illustrations for board games, book illustrations, posters, and concept art, with a programmatic search page ranking terms like "AI fantasy generator" in the top ten. In 2024, a character consistency feature reached number one on Product Hunt, with revenue increasing from about $1 million to approximately $10 million in annual recurring revenue.
The product later shifted to a visual workflow that generates stories from a single sentence, integrating over 50 models covering storyboarding, character continuity, and generating videos and audio, aiming not at a single generator but at connecting scripts to finished pieces. Hollywood reported that its new tool operates as an "atmosphere director": users provide taste and feedback, and the machine produces the output. Meanwhile, Midjourney, Runway, and Sora still focus on models and controls. The team has expanded from ten to several dozen, publicly stating they are hiring for engineering, analytics, business, and content roles.
In market mechanics, buyers are creators and small businesses looking for stable characters and finished pieces, unwilling to manually stitch together multiple models; sellers are OpenArt, which incorporates multiple foundational models into its workflow. The driving force is feature launches and customer acquisition through search, rather than single model releases. Funding from subscription points goes into computing power and workflow development, with the Series A round expanding the team from twenty to over sixty. Beneficiaries include creators who can produce images and videos at low point costs, as well as labs providing foundational models; those under pressure are tools that only sell single generation without character memory and production processes.
Source: Public Information
ABAB AI Insight
Coco Mao's method of validating fit involves deliberately releasing a version without polish. She and John Qiao previously worked on a short video product with a five-person team at Google Area 120, knowing that small teams can generate traffic, but also that product definition rights can disappear once integrated into a platform. In 2022, they started a new venture, initially focusing on prompt browsing rather than developing large models, acknowledging that generation quality would be commoditized by labs, with the real retention being users moving from "seeing an image" to "nurturing a character."
The capital path follows this judgment. The $5 million seed funding came from Basis Set and DCM to set up the search page and points wall; the $30 million Series A was raised after achieving about $70 million in annual recurring revenue and being profitable, aimed at staffing and story workflows, not training foundational models. Each employee generates several million dollars in revenue, indicating that subscriptions still provide substantial margins after subsidizing computing power. The money comes from enthusiasts' monthly fees, which are then paid to model APIs and a few product engineers.
The comparison targets are PhotoRoom and Canva, not Runway. PhotoRoom generates tens of millions in revenue from e-commerce main images; Canva captures edge users of professional software through template workflows. OpenArt applies the same logic to "turning a sentence into a serializable visual IP." The industry phase is packaging layer expansion: model labs compete on image quality and duration, while application layers focus on character stability, editable storyboards, and shareable outputs. Whoever secures the weekly return of creators on that production line will be less likely to be replaced during model upgrades.
Structural changes involve the transfer of pricing power following technological substitution. Substitution occurs at the brush and editing table; pricing power shifts from "whose model is stronger" to "whose workflow allows non-professionals to pay weekly." The mechanism is that lower prices for foundational models will weaken the premium on single generation but raise the entry value of integrating multiple models, character libraries, and publishing formats. The bare-faced test aims to measure whether this entry point still attracts users without brand advertising.
ABAB News · Cognitive Laws
- If people still come without makeup, it indicates they want the product, not the packaging.
- Models will become outdated, but workflows that bring users back weekly will not.
- Earning money before financing keeps pricing power within the product.